JP2011086077A - Storage device, storage system, method of controlling disk array, and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a storage device, a storage system, a method of controlling a disk array, and a program, with which data efficiency can be improved in the storage area of the disk array while suppressing reduction of fault tolerance and performance. <P>SOLUTION: The storage device S-n includes: the disk arrays A-i to A-j; and a disk array control part DE-n which divides the storage area of each disk array to extents and allocates the extents to logical volumes. The disk array control part, if the load of the transfer destination does not satisfy set conditions, specifies the logical volume, collates the logical volume with a policy table in which policy for each logical volume is registered, specifies the policy, and detects a disk array selectable as the transfer destination of the data of the transfer source extent based on the policy. The policy includes the configuration of configurable disk arrays, the priority of data processing, and the logical volumes that could be shared loads. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a storage device that allows a host device to recognize a logical volume, a storage system including the storage device, a method for controlling a disk array in the storage device, and a program.

  In recent years, with the spread of IT, the amount of data used by processing devices and the amount of data processing have increased year by year. Accordingly, the capacity of storage media has been increased, and the processing speed of processing devices has been increased. On the other hand, speeding up of the processing time of a hard disk drive (HDD), which is a magnetic disk, is in a situation where the degree of improvement is low compared to the progress of other processing devices. As a result, the processing time of the HDD becomes a bottleneck of the entire system, which may affect the processing of the entire system.

  Therefore, when operating a system that requires high-speed processing, a high-performance storage device is prepared. Normally, in such a high-performance storage apparatus, a plurality of HDDs constitutes a RAID (Redundant Arrays of Independent Disks). Since the processing to the HDD is distributed by the RAID configuration, it is considered that the speed is increased.

  However, since there is a limit to shortening the seek time and the like due to the movement of the head in the HDD, two or more data that are successively accessed exist at positions apart from each other in the HDD. In this case, there is a problem that the processing takes a long time. In addition, when accesses (hereinafter referred to as “I / O”) occur in a plurality of areas in the same HDD at the same time, these I / Os cannot be processed at the same time until one of the processes ends. There is also a problem that the other process has to wait.

  Under such a background, Patent Document 1 divides the storage area of the HDD into a certain size and optimizes the data stored in the storage area based on the data processing amount per unit in each part. The method of doing is disclosed. According to the method disclosed in Patent Document 1, it is considered that the above problem is solved because the utilization efficiency of the HDD is increased.

  Here, the method disclosed in Patent Document 1 will be specifically described. In the method disclosed in Patent Document 1, first, the storage area of a HDD group (disk array) having a RAID configuration is divided into areas of a certain size (hereinafter referred to as “extents”). Next, an I / O existence rate is obtained for the target HDD or a plurality of extent chunks in the target HDD, and data rearrangement is executed in units of extents based on this I / O existence rate. The

  In the above, an extent chunk refers to an extent configuration area in a single HDD when a certain extent is divided for each HDD. As the I / O presence rate, “simultaneous I / O presence rate”, “extent chunk I / O presence rate”, and “HDD I / O presence rate” are used.

  Among these, the simultaneous I / O existence rate is a probability that a plurality of extent chunks in a certain HDD have I / O being executed at the same time, and is calculated for each extent chunk. Further, in the calculation of the probability that an I / O exists, first, if there is an I / O that is executed at the same time for each sampling for each extent chunk of the representative HDD selected from the HDD array, the counter The value is incremented (+1). Then, the counter value during the sampling period is divided by the number of samplings, and the obtained value becomes the probability that I / O exists, that is, the simultaneous I / O existence rate.

  The extent chunk I / O existence ratio is a probability that an I / O being executed exists in an arbitrary extent chunk in the representative HDD. The extent chunk I / O existence rate is also calculated for each extent chunk. In the calculation of the probability that an I / O exists in this case, first, for each extent chunk of the representative HDD, the counter value is incremented (+1) when one or more I / Os exist at every sampling. . Then, the counter value during the sampling period is divided by the number of samplings, and the obtained value becomes the probability that I / O exists, that is, the extent chunk I / O existence rate.

  Furthermore, the HDD I / O existence ratio is a probability that at least one I / O being executed exists in the representative HDD, and is calculated for each representative HDD. In the calculation of the probability that I / O exists in this case, first, every time sampling, if at least one I / O exists in any extent chunk of the representative HDD, the counter value is incremented (+1). ) Then, the counter value during the sampling period is divided by the number of samplings, and the obtained value is the probability that I / O exists, that is, the HDD I / O existence rate.

  As described above, in the method disclosed in Patent Document 1, data rearrangement is performed mainly in units of extents. The data to be rearranged is rearranged from the extent of the original array to the extent of the other array and optimized, so that the processing speed in the HDD can be increased.

  By the way, when the HDD processes I / O, the processing is performed sequentially as described above. For this reason, even if a plurality of I / O processes are requested at a time, the HDD cannot process them simultaneously, but processes them one by one. Even when there are many I / Os executed in the same area in the HDD, the HDD processes I / O one by one in the same manner.

  For this reason, there is an I / O waiting for processing while the HDD is processing one I / O. At this time, even if the data is rearranged so that the area where many I / Os are transferred to another HDD area, the number of I / Os existing in that area remains the same. An I / O processing wait (hereinafter referred to as an idle state) occurs.

  On the other hand, in the method disclosed in Patent Document 1, data rearrangement for improving the use efficiency of the storage area is based on the frequency of I / O that is executed simultaneously in a plurality of different areas in the same HDD. Done. Further, when the total sum of I / O executed in a plurality of areas in the HDD is large, an idle state occurs as in the case where the load in the same area in the HDD is high. Note that the number of I / Os is not so large when viewed in each area alone.

  Therefore, if the data is rearranged in the HDD having a low frequency of I / O executed simultaneously by the method disclosed in Patent Document 1, the idle state of each I / O is reduced, and the processing speed is increased. The In addition, according to the disclosure method of Patent Document 1, unlike the case where data rearrangement is performed based on simple access frequency to the HDD, effective load distribution is possible. Furthermore, as described above, data can be rearranged based on a small area called an extent, so the load on the rearrangement is reduced, and load balancing at a finer level and adjustment of improvement in access speed can be performed. It becomes executable.

  As described above, the number of I / Os is measured for the representative HDD, whereas the data rearrangement is performed in units of extents, but no particular problem occurs. This is because, basically, the I / O for the HDD group constituting the RAID is striped, so that the number of I / Os in each HDD can be considered to be substantially the same. Therefore, even when one HDD is selected as a representative from the array and measurement is performed on this, a value close to the data obtained when all HDDs are measured can be obtained.

  Furthermore, as described above, I / O is distributed to each HDD, and the actual processing unit of work is considered to be an extent unit. For this reason, even if the data is rearranged in units of extents by the method disclosed in Patent Document 1, no serious problem occurs.

JP 2008-46763 A

  However, the method disclosed in Patent Document 1 has the following problems. The first problem is that only the I / O presence rate is used as a data rearrangement judgment material, and data rearrangement is performed based only on the magnitude of the I / O presence rate. Is a problem that decreases.

  In general, the reason why the RAID configuration is purposely used until the capacity efficiency of the HDD is lowered is to improve the fault tolerance of the stored data in addition to the high-speed processing. The processing speed and the fault tolerance change depending on the type of RAID configured, but the required fault tolerance changes depending on the data used.

  For this reason, the RAID configuration is designed according to the requirements of the data used at the time of initial design. At this time, if it is desired to prevent data loss even if the performance is somewhat degraded, a RAID configuration with high redundancy is adopted even if the performance is somewhat degraded. On the other hand, in the method disclosed in Patent Document 1, since data relocation is not performed in consideration of the design of the RAID configuration, a large problem occurs in the entire system when a failure occurs in the HDD array. As a result, the fault tolerance is reduced.

  The second problem is that important data may be affected when data relocation is executed. Normally, in a system that always requires the highest performance, such as a database for mission-critical business, a configuration that does not use all the capacity of the HDD or a load of important data is used to improve processing performance and avoid occurrence of problems. It may be configured to be arranged in a low region.

  At this time, it is assumed that data of another array is rearranged in such an area because it is not used or the load is low. As a result, when the load on the database of the core business becomes high, the processing of the database may be affected by the processing of the rearranged data. Also in this case, a big problem may occur in the entire system.

  The third problem is that, on the contrary, there is a possibility that the processing performance of the system deteriorates due to the rearrangement of data. In other words, even if data is rearranged to an array that generates a small number of I / Os, if the performance of the HDDs that make up this rearranged array is low, the processing performance of the system decreases. .

  For example, there is a case where data is arranged from a high speed disk such as a SAS (Serial Attached SCSI) disk and a FC (Fibre Channel) disk to a low speed disk such as a SATA (Serial ATA) disk. As described above, when data is rearranged from an HDD with a high rotational speed to an HDD with a low rotational speed, the processing speed may decrease due to the rearrangement.

  The fourth problem is that data to be physically distributed is collected in one place, and as a result, fault tolerance may be reduced. This will be described by taking a database as an example.

  For example, even if the data file that restored the backup data was not the one immediately before the failure when the physical failure occurred in the database, the log file, control file, etc. If there is, it is possible to recover the database to the state immediately before the occurrence of the failure by performing recovery based on them. On the other hand, if there is no log file etc. up to just before the failure occurred, there will be a log file etc. up to just before the failure occurred when recovering the database for the old data file whose backup data was restored Will not. For this reason, it becomes impossible to reflect the update information and the like that existed in the minute to the database, and it is extremely difficult to restore the database to the state immediately before the occurrence of the failure.

  Therefore, in a database, log files, control files, etc. are usually stored in a separate physical area from the data file storage area so that they can be used even if a physical failure occurs. The Also, in the database, log files or the like may be multiplexed as a measure against physical failure. In this case, each file generated by multiplexing is held in a separate storage area physically separated, and fault tolerance is further improved.

  However, when the method disclosed in Patent Document 1 is applied to a database as it is, there is a risk that a data file, a log file, a control file, and the like are held in the same area depending on the load on the database. is there. There is also a similar risk between files generated by multiplexing.

  If these are held in the same area, it becomes difficult to recover the database when a physical failure occurs, and the fault tolerance is reduced. Further, when there is a system that is not limited to a database, and some data should not be stored in the same storage area, the system disclosed in Patent Document 1 is also applied to this system. Fault tolerance is reduced.

  An object of the present invention is to provide a storage device, a storage system, a disk array control method, and a storage device, a storage system, and a disk array control method, which can improve the data efficiency in the storage area of the disk array while solving the above problems and suppressing the fault tolerance and performance degradation To provide a program.

In order to achieve the above object, a storage apparatus according to the present invention is a storage apparatus in which a plurality of logical volumes recognizable from the outside are constructed,
Multiple disk arrays,
A disk array that physically divides a storage area of each of the plurality of disk arrays into a plurality of extents, assigns one or more extents to any of the plurality of logical volumes, and manages the storage areas A control unit,
When the extent of any extent does not satisfy the set condition, the disk array control unit specifies the logical volume to which the transfer source extent is assigned, using the extent as the transfer source extent,
Then, the policy table in which a policy for each of the plurality of logical volumes is registered is compared with the specified logical volume to specify the policy,
Based on the identified policy, a disk array that can be selected as a transfer destination of data stored in the transfer source extent is detected,
The policy for each of the plurality of logical volumes exists in the same disk array as the configuration of the disk array capable of constructing the logical volume, the priority of processing in the data stored in the logical volume, and the logical volume. Including identifiers of other logical volumes that are not allowed,
It is characterized by that.

In order to achieve the above object, a storage system according to the present invention comprises a host device and a storage device in which a plurality of logical volumes recognizable from the host device are constructed,
The storage device physically divides a plurality of disk arrays and a storage area of each of the plurality of disk arrays into a plurality of extents, and allocates one or more extents to any one of the plurality of logical volumes. And a disk array control unit for managing the storage area,
When the extent of any extent does not satisfy the set condition, the disk array control unit specifies the logical volume to which the transfer source extent is assigned, using the extent as the transfer source extent,
Then, the policy table in which a policy for each of the plurality of logical volumes is registered is compared with the specified logical volume to specify the policy,
Based on the identified policy, a disk array that can be selected as a transfer destination of data stored in the transfer source extent is detected,
The policy for each of the plurality of logical volumes includes at least the configuration of a disk array that can construct the logical volume, the priority of processing in the data stored in the logical volume, and the same disk array as the logical volume. Including identifiers of other logical volumes that are not allowed to exist,
It is characterized by that.

In order to achieve the above object, the disk array control method according to the present invention is a method for controlling a plurality of disk arrays provided in a storage apparatus in which a plurality of logical volumes recognizable from the outside are constructed. There,
(A) physically dividing a storage area of each of the plurality of disk arrays into a plurality of extents, and allocating one or two or more extents to any of the plurality of logical volumes;
(B) when the load of any extent does not satisfy the set condition, the logical volume to which the transfer source extent is allocated is specified using the extent as the transfer source extent; and
(C) For each of the plurality of logical volumes, the configuration of the disk array capable of constructing the logical volume, the priority of processing in the data stored in the logical volume, and the same logical disk as the logical volume Collating the logical volume identified in step (b) with a policy table in which a policy is registered, including identifiers of other logical volumes that are not allowed to be identified, and identifying the policy; and
And (d) detecting a disk array that can be selected as a transfer destination of data stored in the transfer source extent based on the policy specified in the step (c). .

Furthermore, in order to achieve the above object, a program in the present invention is a program for controlling a plurality of disk arrays in which a plurality of logical volumes recognizable from the outside are constructed by a computer,
In the computer,
(A) physically dividing a storage area of each of the plurality of disk arrays into a plurality of extents, and allocating one or two or more extents to any of the plurality of logical volumes;
(B) when the load of any extent does not satisfy the set condition, the logical volume to which the transfer source extent is allocated is specified using the extent as the transfer source extent; and
(C) For each of the plurality of logical volumes, the configuration of the disk array capable of constructing the logical volume, the priority of processing in the data stored in the logical volume, and the disk array that is the same as the logical volume Collating the logical volume identified in step (b) with a policy table in which a policy is registered, including identifiers of other logical volumes that are not allowed to be identified, and identifying the policy; and
And (d) detecting a disk array that can be selected as a transfer destination of data stored in the transfer source extent based on the policy specified in the step (c). To do.

  As described above, according to the storage apparatus, storage system, disk array control method, and program of the present invention, the data efficiency in the storage area of the disk array is improved while suppressing the failure resistance and performance degradation. It becomes possible.

FIG. 1 is a block diagram showing a schematic configuration of a storage apparatus and a storage system in an embodiment of the present invention. FIG. 2 is a diagram illustrating disk array management executed in the embodiment of the present invention. FIG. 3 is a diagram showing an example of data rearrangement (extent optimization processing) executed in the embodiment of the present invention. FIG. 4 is a block diagram showing a configuration of the disk array control unit in the embodiment of the present invention. FIG. 5 is a block diagram showing the configuration of the disk controller shown in FIG. FIG. 6 is a diagram showing an example of a policy table used in the embodiment of the present invention. FIG. 7 is a diagram illustrating processing based on processing priority. FIG. 8 is a diagram illustrating an example of data arrangement for avoiding a disk failure. FIG. 9 is a diagram showing an example of a RAID correspondence table used in the embodiment of the present invention. FIG. 10 is a flowchart showing processing executed by the disk array control unit constituting the storage apparatus according to the embodiment of the present invention. FIG. 11 is a flowchart showing processing of the disk array control unit after step F9 shown in FIG. FIG. 12 is a flowchart showing error processing executed by the disk array control unit constituting the storage apparatus according to the embodiment of the present invention. FIG. 13 is a diagram for explaining a modification of the embodiment of the present invention. FIG. 14 is a flowchart showing error processing executed by the disk array control unit constituting the storage apparatus in the modification of the embodiment of the present invention.

(Summary of Invention)
The present invention is applied to a storage apparatus. In the present invention, the storage area of a disk array composed of a plurality of HDDs is regarded as one continuous area, and is managed by being divided into areas (extents) of a certain size. Furthermore, an extent configuration area in a single HDD when an extent is divided for each HDD is called an extent chunk.

  In the present invention, as in the invention disclosed in Patent Document 1 shown in the background art section, when the load of any extent set in the storage area of the disk array becomes high, the extent is stored. Stored data is transferred and relocated to other less loaded extents.

  Specifically, for example, in a certain extent chunk, when the I / O existence rate acquired from the disk array is higher than a preset reference value, the I / O existence rates of other extent chunks are referred to. Is done. Then, extent chunks having an I / O presence rate lower than the reference value are detected, and a transfer destination extent chunk is selected from the extent chunks. Thereafter, data exchange (data migration) is performed between extent chunks having an I / O presence rate higher than a reference value and extent chunks having an I / O presence rate lower than a reference value.

  Also in the present invention, as in the invention disclosed in Patent Document 1, as the I / O existence ratio serving as a determination index, the above-mentioned “simultaneous I / O existence ratio” and “extent chunk I / O existence ratio” are used. And “HDD I / O presence rate” can be used. As described above, the present invention has portions in common with the invention disclosed in Patent Document 1.

  However, in the present invention, when determining a transfer destination extent (extent chunk), unlike the invention disclosed in Patent Document 1, a policy table is used as another determination index. In addition, a RAID correspondence table, a chunk correspondence table, and an extent correspondence table can also be used.

  The policy correspondence table (see FIG. 6 described later) registers a policy for each logical volume recognized by the host device (server group) connected to the storage device. Here, a logical volume refers to a logical unit (LU) constructed by logically dividing a disk array. Each logical unit is assigned a logical unit number (LUN) for identification. Has been.

  In addition, as registered policies, for example, processing performance and fault tolerance (RAID configuration) of a disk array capable of arranging data for each logical volume (LU), processing of the entire storage device for each logical volume (LU) A priority etc. are mentioned.

  Further, the RAID correspondence table (see FIG. 9 described later) registers the type and performance of the RAID configuration for each disk array provided in the storage apparatus. The chunk correspondence table links extents and extent chunks corresponding to the extents. With the chunk correspondence table, it is possible to confirm to which extent each extent chunk belongs.

  The extent correspondence table associates a logical extent (LE) in a logical volume with a corresponding physical extent (PE). According to the extent correspondence table, it is possible to confirm which physical extent on the disk array corresponds to the logical extent of each logical volume. In this specification, when “extent” is simply written, it means “physical extent”. As described above, when it is necessary to distinguish between a physical extent and a logical extent, they are represented as “logical extent” or “physical extent”.

  Here, an example of processing in the present invention will be specifically described. First, as a premise, it is assumed that the load of any extent chunk set in the storage area of the disk array is high. Next, the chunk correspondence table is referred to, and the extent to which the extent chunk having a high load belongs is specified. Data stored in the specified extent is a target for transfer (relocation). The specified extent is hereinafter referred to as “transfer source extent”. Next, the extent correspondence table is referred to, and the logical volume (LUN) to which the transfer source extent is allocated is searched.

  If the assigned logical volume (LUN) is found as a result of the search, a policy correspondence table created in advance is referred to and the policy of the found logical volume (LUN) is obtained. Next, the obtained policy is applied to the RAID correspondence table, and a determination is made as to whether or not there is a disk array to which data can be rearranged.

  As a result of the determination, if there is no disk array that conforms to the policy, data rearrangement is not performed. If the data is not rearranged, the process ends. In addition, when processing is completed without relocation, the fact that relocation could not be executed and the reason why it could not be executed are described for the management terminal connected to the storage device Error message is sent. Note that, in the subsequent operation, when the processing is completed without executing the rearrangement, a similar error message is transmitted to the management terminal device.

  On the other hand, if there is a disk array that conforms to the policy as a result of the determination, the extent load status of each disk array that can be rearranged is checked. As a result of the investigation, if the load is greater than or equal to the specified value in all areas of the disk array, data rearrangement is not performed, and the process ends. On the other hand, if the result of the investigation shows that the load is below the specified value in any storage area of the disk array, the following processing is subsequently performed.

  In this case, first, the storage area with the lowest load is detected in the storage area of the disk array. Subsequently, the extent chunk with the lowest load is detected in the detected storage area. Then, referring to the chunk correspondence table, the extent to which the detected extent chunk belongs is detected. The detected extent is an extent that is a transfer destination candidate (transfer destination candidate extent).

  Subsequently, the extent correspondence table is referenced to detect the logical volume (LUN) to which the transfer destination candidate extent is allocated. Further, referring to the policy correspondence table, the policy of the logical volume to which the detected extent is allocated is specified.

  If the transfer destination candidate extent is not allocated to any logical volume, that is, is an unused area, there is no policy corresponding to the extent. Therefore, in that case, a policy set exclusively for the unused area can be used.

  Next, when the policy of the logical volume to which the transfer destination candidate extent is assigned is found, the found policy is compared with the policy of the logical volume to which the transfer source extent assigned earlier is assigned. .

  In the above comparison, for example, whether or not the data of the transfer source extent can be relocated to the disk array in which the transfer destination candidate extent exists is examined. Then, it is determined whether the data of the transfer source extent can be transferred to the extent chunk having the lowest load of the transfer destination candidate extent.

  As a result of the comparison, if relocation by data transfer is impossible, the same determination is performed again for the extent chunk with the next lowest load. At this time, extent chunks determined to be impossible to be rearranged by data transfer are excluded from the determination.

  On the other hand, as a result of the comparison, when relocation by data transfer becomes possible, all the allocated logical volumes are specified for the area of the disk array in which this transfer destination candidate extent is set. The logical volume is specified by using the method described above by using the chunk correspondence table, extent correspondence table, and policy correspondence table.

  When the policies are found for all the allocated logical volumes, the respective policies are compared with the load status of the transfer source extent. For example, in this comparison, it is examined whether processing in a logical volume that already uses a disk array area is not under pressure.

  As a result of the comparison, if relocation by data transfer is impossible, it is impossible to transfer data to the disk array of the transfer destination candidate extent that is the comparison target. Identify extent chunks. Then, the above-described processes such as specifying the policy of the corresponding logical volume and comparing the policies are performed again.

  On the other hand, when rearrangement by data transfer is possible, data transfer is performed by a module or the like that executes data rearrangement, and the rearrangement is executed. Then, after the rearrangement is completed, the extent table and the chunk correspondence table are updated.

  If relocation by data transfer is not possible in all areas of the disk array in which the transfer destination candidate extent exists, data relocation is not performed and the process ends. By performing such processing, it is possible to prevent adverse effects such as a decrease in fault tolerance of the disk array and pressure on processing to be prioritized due to data rearrangement.

(Comparison between the present invention and the invention disclosed in Patent Document 1)
In the invention disclosed in Patent Document 1, the data in an extent does not necessarily belong to the same logical volume. On the other hand, in the present invention, data rearrangement is executed by using the logical volume policy, so that all data in one extent can belong to the same logical volume.

  Also, the data in the extent can be assigned to a single logical volume as follows. First, when writing to each logical volume occurs, I / O to the disk array is performed sequentially (that is, one by one). Therefore, for example, the writing to each logical volume may be collected for each logical volume on the storage device cache until the specified value is reached, and written to the HDD after reaching the specified value.

  Although details will be described later, in the present invention, data rearrangement can be executed between disk arrays having different RAID configurations. However, since the number of HDDs differs depending on the RAID configuration that constitutes the disk array, and the extent capacity changes, if the capacity of data transferred in units of extent chunks is set, the data can be regenerated between extents. Placement becomes difficult.

  Therefore, in the present invention, the capacity of data to be transferred can be set in units of extents. As a result, the amount of data allocated to each HDD is different for each disk array, but the extent data amount is the same for all disk arrays.

(Embodiment)
Hereinafter, a storage apparatus, a storage system, a disk array control method, and a program according to an embodiment of the present invention will be described with reference to FIGS. First, the configuration of the storage apparatus and the storage system in this embodiment will be described with reference to FIG. FIG. 1 is a block diagram showing a schematic configuration of a storage apparatus and a storage system in an embodiment of the present invention.

  As shown in FIG. 1, the storage system 1 is an autonomous data rearrangement system, and includes a host device 10 and storage devices S-1 and S-2. In each of the storage devices S-1 and S-2, a plurality of logical volumes recognizable from the host device 10 are constructed. The host device 10, the storage device S-1, and the storage device S-2 are connected to each other via the network 20.

  In the present embodiment, only two storage devices S-1 and S-2 are shown in FIG. 1, but the number of storage devices is not particularly limited. The storage system 1 only needs to include one or more storage devices. Specific examples of the network 20 include Fiber Channel (FC) and Internet SCSI (iSCSI).

  Further, as shown in FIG. 1, each of the storage apparatuses S-1 and S-2 includes a plurality of disk arrays A-i to A-j and a disk array control unit DE-n. The disk array control unit DE-n is an array controller that accesses each disk array in response to a request from the host device 10. Note that i and j are natural numbers, and i <j. n is a natural number and corresponds to a number assigned as a code to the storage device.

  Further, in the present embodiment, as shown in FIG. 1, the storage apparatus S-1 includes three disk arrays A-1, A-2, and A-3. On the other hand, the storage apparatus S-2 includes three disk arrays A-4, A-5, and A-6. The disk arrays A-1, A-2, A-3, A-4, A-5, and A-6 have identification numbers # 1, # 2, # 3, # 4, # 5, and # 6, respectively. Is granted. In each storage apparatus, the number of disk arrays may be plural, and is not particularly limited. In the following description, the disk array is simply referred to as “array”.

  Each array only needs to include at least one HDD (disk device), and the number of HDDs included in each array is not particularly limited. In this embodiment, as shown in FIG. 1, the arrays A-1 and A-2 of the storage device S-1 and the arrays A-4 and A-5 of the storage device S-2 are, for example, four HDDs 100. It has. The arrays A-3 and A-6 include three HDDs 100. Further, although not shown, in the storage apparatuses S-1 and S-2, each HDD 100 is assigned a unique identification number (HDD number) in the storage apparatuses S-1 and S-2.

  In this embodiment, each of the arrays A-1 to A-6 has a RAID configuration to which, for example, RAID5 or RAID10 is applied. The storage area of each array is constructed by combining the HDDs 100 that constitute the array A-i, and is defined by taking the storage area of each HDD 100 as one continuous storage area.

  In the storage apparatus S-1, the disk array control unit DE-1 physically divides each storage area of the arrays A-1 to A-3 into a plurality of extents, and either one or two or more extents. Assign to one of these logical volumes. The disk array control unit DE-1 manages the storage areas of the arrays A-1 to A-3 in extent units.

  Similarly, in the storage apparatus S-2, the disk array control unit DE-2 physically divides each storage area of the arrays A-4 to A-6 into a plurality of extents, and one or more extents are allocated. Assign to one of the logical volumes. The disk array control unit DE-2 also manages the storage areas of the arrays A-4 to A-6 in extent units.

  Specifically, the extents that the disk array control units DE-1 and DE-2 set by physically dividing the storage area of the array are physical extents. The disk array control units DE-1 and DE-2 assign each physical extent to each logical extent that constitutes the logical volume. Note that the number of logical extents matches the number of physical extents.

  In the present embodiment, the logical extent and the physical extent are associated with each other by the extent table. The extent table is managed by a mapping table. Here, it is assumed that the logical extent is LE, and each LE is assigned an identification number such as LE1, LE2,. Similarly, it is assumed that the physical extent is a PE, and each PE is assigned an identification number such as PE1, PE2,.

  For example, assume that the PE associated with LE1 is changed from PE1 to PE2. In this case, in the mapping table, the mapping information indicating the correspondence between LE1 and PE1 is updated to the mapping information indicating the correspondence between LE1 and PE2. As a result, the extent table is also updated. Although already described, PE is simply expressed as “extent” unless it is necessary to distinguish between PE and LE.

  Also, the disk array control units DE-1 and DE-2 are logical volumes to which a transfer source extent is assigned using the extent as a transfer source extent when the load of any extent does not satisfy the set condition. Is identified. Then, the disk array control units DE-1 and DE-2 identify the policy by collating the identified logical volume with a policy table (see FIG. 6 described later) in which a policy for each logical volume is registered. . Thereafter, the disk array control units DE-1 and DE-2 detect a disk array that can be selected as a transfer destination of data stored in the transfer source extent based on the identified policy.

  The policy for each logical volume is permitted to exist in the same array as the logical volume, the configuration of the disk array in which the logical volume can be constructed, the processing priority of the data stored in the logical volume, and so on. It contains at least the identifiers of other logical volumes that are not. The logical volume identifier means the logical unit number (LUN) described above.

  As described above, in this embodiment, when the load of any extent becomes high and data transfer becomes necessary, the policy of the logical volume to which the transfer source extent is allocated is specified, and the policy is based on the policy. The destination array is selected. Since the policy includes a configuration required for the array at the transfer destination, a reduction in fault tolerance is suppressed. In addition, since the priority of data processing is included, a decrease in processing performance at the transfer destination is also suppressed. Furthermore, since the policy includes LUNs of logical volumes that cannot be constructed in the same array, it is avoided that data to be physically stored in separate areas is stored in the same array. A decrease in fault tolerance is suppressed.

  From these points, according to the storage devices DE-1 and DE-2 and the storage system 1, it is possible to improve the data efficiency in the storage area of the array while suppressing the failure tolerance and the performance degradation. It becomes.

  Here, the configurations of the storage device DE-1, the storage device DE-2, and the storage system 1 will be described more specifically with reference to FIGS. First, the management using the extent of the storage area of the disk array will be described with reference to FIG. FIG. 2 is a diagram illustrating disk array management executed in the embodiment of the present invention.

  As shown in FIG. 2, the storage area of the array composed of a plurality of HDDs 100 is divided into a plurality of extents and managed. In addition, when an extent is divided for each HDD, the extent configuration area in the single HDD is an extent chunk.

  As described above, the disk array control units DE-1 and DE-2 access the logical volume constructed in the array A-i in response to an instruction from the host device 10. Then, the disk array control units DE-1 and DE-2 manage the correspondence relationship between the logical volume and the LEs constituting the logical volume and the correspondence relationship between each LE and the PE to which the LE is mapped, using the extent table. .

  In addition, the disk array control units DE-1 and DE-2 assign specific extent chunks in the storage device SE-n by combining the extent number assigned to each extent and the HDD number assigned to each HDD. Can be recognized uniquely.

  Further, the disk array control units DE-1 and DE-2 execute autonomous rearrangement of data. In other words, the disk array control units DE-1 and DE-2 are connected between arrays in a storage device supported by the disk array control unit, or between storage device arrays supported by other disk array control units ( Move data (across storage devices).

  Specifically, the disk array control units DE-1 and DE-2 copy data of two extents, and the storage destination PE of the copied data is stored within the storage apparatus or across the storage apparatuses. And replace these copied data.

  Such data rearrangement is generally called optimization. However, in the present invention, since extents in which data is stored are replaced and data rearrangement is performed, this is referred to as “extent optimization”.

  Next, referring to FIG. 3, an example in which the disk array control unit DE-1 or DE-2 executes the above extent optimization between an extent in one array and an extent in another array. explain. FIG. 3 is a diagram showing an example of data rearrangement (extent optimization processing) executed in the embodiment of the present invention.

  As shown in FIG. 3, in the array (# X1), there is an extent whose simultaneous I / O existence rate is higher than a preset reference value. Of the extent chunks constituting this extent, the extent chunk A of the HDD 100a constituting the array (# X1) has the highest degree of existence of the waiting I / O. That is, extent chunk A is the extent chunk with the highest load in HDD 100a.

  Also, as shown in FIG. 3, in the array (# X2), there are many extents whose simultaneous I / O existence ratio is lower than a preset reference value. Of the extent chunks constituting this extent, the extent chunk B of the HDD 100b constituting the array (# X2) has the lowest degree of existence of the waiting I / O. That is, extent chunk B is the extent chunk with the lowest load in HDD 100b.

  In the example of FIG. 3, extent optimization is performed between two extents of extent chunk A and extent chunk B. As a result, in the HDD 100a and the HDDb, the use efficiency of the storage area is improved, and the processing performance in the HDD is also improved.

  Next, the configuration of the disk array control unit will be specifically described with reference to FIG. FIG. 4 is a block diagram showing a configuration of the disk array control unit in the embodiment of the present invention. However, in this embodiment, the disk array control units DE-1 and DE-2 have the same configuration. Furthermore, in this embodiment, when the number of storage devices increases, the number of disk array control units having the same configuration as the disk array control units DE-1 and DE-2 increases. Therefore, in the following description, the disk array control unit DE-n is taken as an example. Note that n is a natural number as described above.

  As shown in FIG. 4, in this embodiment, the disk array controller DE-n includes a host controller C1, a main controller C2, a disk controller C3, a cache memory C4, and a memory C5.

  The host controller C1 is connected to the host device 10 (see FIG. 1) via the network 20 (not shown in FIG. 4), and executes communication processing with the host device 10. Specifically, the host controller C1 has a port for transmitting and receiving data to and from the host device 10, and a controller (driver) that manages and processes data and commands.

  The main controller C2 manages the entire disk array control unit DE-n. Specifically, the main controller C2 controls storage functions such as cache control and disk replication.

  The cache C4 temporarily stores each data processed by the main controller C2, for example, data generated by I / O between the host device 10 and each array, and accelerates the processing speed. .

  The disk controller C3 performs communication processing with each HDD 100 constituting the array and executes optimization processing. Specifically, the disk controller C3 performs RAID control, control of connection ports connected to each HDD 100, and the like. The disk controller C3 also manages each table such as an extent correspondence table and a chunk correspondence table.

  Further, the disk controller C3 detects the standby I / O of each HDD, and calculates the above-described simultaneous I / O presence rate, HDD I / O presence rate, and extent chunk I / O presence rate. Further, the disk controller C3 performs the extent optimization described above based on these existence rates. The processing by the disk controller C3 is performed on the memory 5C. The memory C5 stores data handled in the optimization process executed by the disk controller C3.

  In the present embodiment, the disk array control unit DE-n can be realized by causing a computer to execute a program in the present embodiment to be described later. In this case, a CPU (Central Processing Unit) of the computer functions as a host controller C1, a main controller C2, and a disk controller C3. Further, the memory provided in the computer functions as the cache memory C4 and the memory C5.

  Next, the configuration and functions of the disk controller C3 shown in FIG. 4 will be specifically described with reference to FIGS. FIG. 5 is a block diagram showing the configuration of the disk controller shown in FIG.

  As shown in FIG. 5, the disk controller C3 includes an autonomous optimization control engine D2, an I / O information collection module D3, a chunk correspondence table D4, an extent correspondence table D5, a policy correspondence table D6, and a RAID correspondence table. D7 and an extent relocation execution module D8. Further, the disk controller C3 is connected to an input module D1 that receives data input from the outside, a memory C5, and each HDD 100 that constitutes the array.

  Although not shown, each HDD 100 is provided with an HDD driver that measures the number of I / Os generated in the storage area. The HDD driver stores the measured number of I / Os in the memory C5. The I / O information collection module D3 collects the number of I / Os of each HDD stored in the memory C5, performs sampling at a set cycle, and calculates the I / O existence rate.

  The autonomous optimization control engine D2 determines whether or not extent optimization is necessary based on the I / O presence rate calculated by the I / O information collection module D3. As a result of the determination, if necessary, the independent optimization control engine D2 refers to the chunk correspondence table D4, the extent correspondence table D5, the policy correspondence table D6, and the RAID correspondence table D7, and determines whether extent optimization is possible. To do.

  When it is determined that extent optimization is possible, the autonomous optimization control engine D2 passes information on extent optimization to the extent relocation execution module D8. The extent rearrangement execution module D8 executes extent optimization based on the passed information.

  In the present embodiment, the reference value used for determination by the autonomous optimization control engine D2 and the policy correspondence table D6 are input by the input module D1. The input module D1 is a module that exchanges data with the outside in the disarray control unit DE-n, and can be realized by application software such as storage management software, for example.

  Further, among the tables used in the present embodiment, the chunk correspondence table D4 and the extent correspondence table D5 are realized by a mapping table. The policy correspondence table D6 will be described with reference to FIG. The RAID correspondence table D7 will be described with reference to FIG.

  FIG. 6 is a diagram showing an example of a policy table used in the embodiment of the present invention. As shown in FIG. 6, a policy for each logical volume is registered in the policy correspondence table. In the example of FIG. 6, each logical volume is distinguished by a LUN (logical unit number). As described above, the LUN is a number uniquely assigned to the logical volume in the storage apparatus, and each logical volume can be identified by the LUN. In addition, “allowable array configuration”, “allowable performance”, “processing priority”, “non-sharable LUN”, and “allowable work load level” are described as policies.

  The allowed array configuration indicates a configuration of a disk array that can construct a corresponding logical volume, specifically, a RAID configuration. In other words, if the array configuration described in the allowable array configuration is used, the data of the corresponding logical volume can be held. If an array having a configuration described as an allowable array configuration is selected as the transfer destination array, fault tolerance (redundancy) and performance can be ensured after rearrangement. Further, by not selecting an array having a configuration other than the described configuration as a transfer destination, it is possible to prevent fault tolerance (redundancy) and performance degradation due to optimization.

  “Allowable performance” indicates an index indicating the performance of the array capable of constructing the corresponding logical volume. The performance of the array is also judged by “allowable array configuration”, but the performance of the array is judged more finely by “allowable performance”. Furthermore, by referring to the index described as the allowable performance, it can be determined whether or not the data of the extent of the transfer source can be rearranged in the transfer destination array.

  If an array with performance exceeding the level indicated in the permissible performance is selected as the destination array and extent optimization is performed, performance degradation due to optimization can be prevented more reliably. . Specifically, if the allowable performance of the LUN row corresponding to the transfer source extent is B, an array having an allowable performance of B or higher, that is, an array of A or B may be selected as the transfer destination array. . In the example of FIG. 6, the performance increases in the order of A, B, and C. In addition, hereinafter, even when the same index is used, it is assumed that the lower order is shown in the same order.

  “Processing priority” indicates the processing priority in the data stored in the corresponding logical volume. In other words, the processing priority indicates the priority of the business using the logical volume of the target LUN. By referring to the processing priority, the processing priority can be compared between the transfer destination extent and the transfer source extent. Then, as a result of the comparison, when the priority of the transfer source extent is lower than the priority of the transfer destination extent, it is preferable not to execute extent optimization.

  Here, the priority of processing will be described with reference to FIG. FIG. 7 is a diagram illustrating processing based on processing priority. Specifically, as shown in FIG. 7, it is assumed that the priority of processing of the logical volume Y2 to which the transfer source extent is allocated is B. In this case, it is preferable not to select an extent allocated to a logical volume whose processing priority is A or higher as the transfer destination array.

  In other words, not the extent to which the logical volume Y1 with the priority A is allocated, but the extent to which the logical volume Y3 with the priority B is allocated is selected, and data rearrangement (extent optimization) is performed on this extent. Is done. As a result, it is possible to avoid a situation in which an excessive load is applied to the processing of important work.

  “Non-sharable LUN” indicates an identifier of another logical volume that is not allowed to exist in the same array as the corresponding logical volume. That is, the LUN that cannot be arranged in the same array is described in the non-sharable LUN. For example, among the data used, there are cases where two pieces of data should not exist on the same disk assuming a disk failure. This point will be described with reference to FIG. FIG. 8 is a diagram illustrating an example of data arrangement for avoiding a disk failure.

  As shown in FIG. 8, it is assumed that data of the database server H-1, the file server H-2, and the file server H-3 are stored in the storage device Sn. Since the data file, log 1 and log 2 used by the database server H-1 cannot be stored in the same array, they are stored in the arrays L-1, L-2 and L-3, respectively. Yes. This is because, in general, log files are made redundant in databases, but in order to ensure fault tolerance, redundant log files and data files usually need to be placed in separate arrays. It is.

  Also, the file area A and the file area B used by the file server H-2 are data that cannot be stored in the same array, and are therefore stored in the arrays L-4 and L-5, respectively. Furthermore, since the file area C and the file area D used by the file server H-3 cannot be stored in the same array, they are stored in the arrays L-6 and L-7, respectively.

  “Unshareable LUN” prevents data that cannot be stored in the same array shown in FIG. 7 from being stored in the same array by extent optimization. That is, when viewed from the storage device, the disk failure is determined in units of arrays. Therefore, when this unshareable LUN is referred to and the LUN described here is in the relocation destination, the optimization is not executed. This avoids a situation in which logical volumes that should not be in the same array are placed in the same array due to extent optimization. As a result, a reduction in the fault tolerance of the system is prevented.

  “Allowable work load level” indicates an allowable load that an array in which a corresponding logical volume is constructed can be allowed for another logical volume. In other words, the allowable workload level describes the allowable load level of extents transferred from other arrays. The allowable workload level is used when one extent is not subject to optimization, but another extent configured in the same array is subject to optimization, and processing in the former extent is optimized. Suppresses being pressed.

  In other words, if an extent with a high load is transferred, the processing of the existing extent may be under pressure. For this reason, when the load of the transfer source extent is larger than the allowable work load level of the transfer destination logical volume due to the allowable work load level, optimization is not performed. As a result, a situation in which processing of important work is compressed is avoided.

  For example, it is assumed that the allowable work load level of the logical volume to which the transfer source extent is allocated is B. In this case, if a logical volume of another extent having the same array as the transfer destination extent is specified and the allowable work load level of the logical volume is C, relocation is executed.

  Thus, the “allowable work load level” is an index different from the “processing priority” described above. For example, it is assumed that there is a logical volume arranged in an array in which a logical volume with a small load is arranged in order to prevent a decrease in processing performance. In such a case, if an extent with a high load is rearranged in this array, there is a possibility that the processing of the logical volume arranged so as not to degrade the processing performance is compressed.

  With “processing priority” alone, it is difficult to avoid the above-mentioned pressure because optimization is performed on extents with low importance and unused areas of the array. When “level” is used, it is easy to avoid the above compression. Therefore, by using the “allowable work load level”, an extra load is not applied to the work of the logical volume other than the optimization target, and a finer control of the work load becomes possible.

  In the storage area of the array, there is also a storage area (unused area) in which no logical volume is constructed. As shown in FIG. 6, a policy for an unused area may be registered in the policy correspondence table. It should be noted that any extent may be rearranged so that there is no restriction on each item of the unused area, and a restriction may be set if necessary.

  FIG. 9 is a diagram showing an example of a RAID correspondence table used in the embodiment of the present invention. As shown in FIG. 9, the RAID correspondence table registers the type of RAID configuration of the array, the performance of the array, and the number of constituent HDDs for each array number. The array number is an identification number for identifying each array in the RAID correspondence table, and is uniquely assigned to the array in the storage apparatus. Each array is identified by the array number. Note that the performance of the array is represented by an index.

  Next, the operation of the disk array control unit DE-n described above will be described with reference to FIGS. FIG. 10 is a flowchart showing processing executed by the disk array control unit constituting the storage apparatus according to the embodiment of the present invention. FIG. 11 is a flowchart showing processing of the disk array control unit after step F9 shown in FIG.

  In the present embodiment, the disk array control method is implemented by operating the disk array control unit DE-n. Therefore, the description of the disk array control method in the present embodiment is replaced with the following description of the operation of the disk array control unit DE-n. In addition, FIGS. 1 to 9 are referred to as appropriate.

  As shown in FIG. 10, first, in the disk array control unit DE-n, the autonomous optimization control engine D2 of the disk controller C3 determines whether or not there is an extent chunk whose load is higher than the reference value (step F1). . Note that the determination in Step F1 can be performed based on the I / O existence ratio, as in the invention disclosed in Patent Document 1.

  If the result of determination in step F1 is that there is no extent chunk having a higher load than the reference value, the processing in the disk array control unit DE-n ends. On the other hand, if the result of determination in step F1 is that there is an extent chunk with a higher load than the reference value, the autonomous optimization control engine D2 detects an extent chunk with a higher load to be optimized (step F2).

  Next, when detecting an extent chunk with a high load, the autonomous optimization control engine D2 refers to the chunk correspondence table D4 and detects the extent to which the extent chunk detected in step F2 belongs (step F3). Next, when the extent to which the autonomous optimization control engine D2 is found is found, the autonomous optimization control engine D2 refers to the extent correspondence table D5 and detects the LUN of the logical volume to which the extent belongs (step F4).

  Next, when detecting the LUN, the autonomous optimization control engine D2 refers to the policy correspondence table (see FIG. 6), and detects a policy corresponding to the detected LUN (step F5). Furthermore, when detecting the policy, the autonomous optimization control engine D2 refers to the RAID correspondence table D7 (see FIG. 9) and detects an array that can correspond to the detected policy (step F6).

  Next, the autonomous optimization control engine D2 determines whether there is an array that can be selected as a data transfer destination in the extent chunk detected in Step F2 among the arrays detected in Step F6 (Step S6). F7). Specifically, in step F7, “allowable array configuration” and “allowable performance” in the policy correspondence table (see FIG. 6) and “RAID configuration” and “array performance” in the RAID correspondence table (see FIG. 9) are obtained. To be compared. Then, from the comparison result, the autonomous optimization control engine D2 determines whether the array is selectable as a data transfer destination (transferable array).

  If the result of determination in step F7 is that there is no array that can be selected as the transfer destination, the processing in the disk array control unit DE-n ends. On the other hand, if there is an array that can be selected as the transfer destination, the autonomous optimization control engine D2 acquires the I / O existence rate calculated by the I / O information collection module D3 and determines that the transfer is possible in step F7. The load status of the storage area of the array thus detected is detected (step F8).

  Next, the autonomous optimization control engine D2 uses the acquired I / O existence rate, and among the arrays determined to be selectable in step F7, there is an array whose storage area load is equal to or less than the reference value. (Step F9).

  As a result of the determination in step F9, when there is no array whose storage area load is equal to or less than the reference value, the processing in the disk array control unit DE-n is ended. On the other hand, if the result of determination in step F9 is that there is an array in which the load on the storage area is below the reference value, the autonomous optimization control engine D2 executes step F10 (see FIG. 11).

  As shown in FIG. 11, in step F10, the autonomous optimization control engine D2 detects an array with the lowest load in the storage area. The autonomous optimization control engine D2 also performs detection using the I / O presence rate calculated by the I / O information collection module D3 in step F10.

  Next, when the autonomous optimization control engine D2 detects an array having the lowest load in the storage area, the autonomous optimization control engine D2 detects an extent chunk having the lowest load in the storage area of the detected array (step F11). Subsequently, the autonomous optimization control engine D2 executes extent detection (step F12), LUN detection (step F13), and policy detection corresponding to the LUN (step F14). Steps F12, F13, and F14 are the same as steps F3, F4, and F5, respectively.

  Next, the autonomous optimization control engine D2 compares the transfer source LUN policy detected in step F5 with the transfer destination candidate LUN policy detected in step F14 (step F15). Next, the autonomous optimization control engine D2 determines whether or not the transfer source data can be transferred to the extent detected in step F12 and rearranged based on the comparison result (step F16).

  Specifically, in step F15, “allowable array configuration”, “allowable performance”, “processing priority”, and “non-sharable LUN” in the policy correspondence table (see FIG. 6) are referred to for comparison. In step F16, it is first determined whether the transfer destination candidate array can accommodate the transfer source extent from the viewpoints of “allowable array configuration” and “allowable performance”.

  If it is determined in the above-described determination in step F16 that it is not possible to handle, processing is performed assuming that relocation is impossible. On the other hand, if it is determined that it can be handled, it is further determined whether or not the transfer destination candidate array can support the transfer source extent in terms of “processing priority” and “shared load LUN”.

  Then, if the priority corresponding to the extent of the transfer destination is higher than the priority corresponding to the extent of the transfer source, processing is performed assuming that relocation is impossible. In addition, if the logical volume to which the transfer source extent is assigned and the logical volume to which the transfer destination extent is assigned cannot be constructed in the storage area of the same array, the relocation is not possible. The processing is performed assuming that

  Next, when the rearrangement is impossible as a result of the determination in step F16, the autonomous optimization control engine D2 detects the extent chunk having the next lowest load from the storage area of the array detected in step F10. (Step F17). At this time, in step F17, the extent chunk to which the logical volume that is the comparison target of the policy in step F15 is allocated (the extent chunk detected in the previous step F11) is excluded from the detection target.

  Next, the autonomous optimization control engine D2 determines whether there is a logical volume to which the extent chunk detected in step F17 is allocated (step F18). That is, in step F18, it is determined whether or not a logical volume (LU) that can be used as a transfer destination exists in the storage area of the array detected in the previous step F8. If the result of determination in step F18 is that a logical volume exists, step F11 and subsequent steps are executed again. On the other hand, when the logical volume does not exist, Step F23 described later is executed.

  If the result of determination in step F16 shows that relocation is possible, the autonomous optimization control engine D2 refers to the extent correspondence table D5 and detects all logical volumes built in the array that is the transfer destination candidate. (Step F19). In step F19, the transfer destination candidate logical volume used in step F15 is excluded. For this reason, in step F21 to be described later, it is possible to suppress a useless process in which a comparison is performed again for a logical volume that has already been compared.

  Next, the autonomous optimization control engine D2 refers to the policy correspondence table and detects the policy of each logical volume detected in step F19 (step F20). Subsequently, the autonomous optimization control engine D2 compares all the policies detected in step F20 with the load of the transfer source extent (step F21). Further, the autonomous optimization control engine D2 determines whether relocation by relocation is possible based on the comparison result in step F21 (step F22).

  Specifically, in step F21, the comparison is performed using the “non-shareable LUN” and the “allowable work load level” in the policy correspondence table (see FIG. 6). In step F22, if there is a logical volume that cannot be shared due to the relationship with the logical volume to which the transfer source extent is allocated, the processing is performed assuming that relocation is impossible. Furthermore, in step F22, even when the load of the transfer source extent is larger than the allowable work load level of the logical volume constructed in the transfer destination candidate array, the processing is performed as being impossible to relocate.

  If there is no LUN that cannot be shared and the load is also at an allowable level, the processing is performed as relocation possible. If not possible, step F23 is performed. In step F23, it is determined whether there is any other relocatable array. If there is no array, the process ends.

  If transfer is possible in the determination in step F22, the autonomous optimization control engine D2 passes information on data rearrangement (extent optimization) to the extent rearrangement execution module D8. As a result, the extent rearrangement execution module D8 executes data rearrangement (extent optimization) (step F25). Thereafter, the processing in the disk array control unit DE-n ends.

  On the other hand, when transfer is not possible in the determination in step F22, the autonomous optimization control engine D2 executes step F23. In step F23, the autonomous optimization control engine D2 refers to the result of step F9 and determines whether there is another array that can be selected as a data transfer destination.

  If there is another array as a result of the determination in step F23, the autonomous optimization control engine D2 selects this so that the array detected in the latest step F10 does not become a data transfer destination candidate. (Step F24). Then, the autonomous optimization control engine D2 executes the process from Step F10 again.

  When data transfer is possible by executing Steps F1 to F24 shown in FIGS. 10 and 11, rearrangement by data transfer is executed, and extent optimization is performed. On the other hand, when the data transfer is impossible, the rearrangement by the data transfer is not executed, and the processing in the disk array control unit DE-n is finished.

  Here, in the processing shown in FIG. 10 or FIG. 11, processing in the storage apparatus when the processing ends without transferring data will be described with reference to FIG. 12. FIG. 12 is a flowchart showing error processing executed by the disk array control unit constituting the storage apparatus according to the embodiment of the present invention, and shows error processing.

  As shown in FIG. 12, first, in the disk array control unit DE-n, when the main controller C2 finishes the process without performing the data rearrangement, it indicates that the data rearrangement could not be performed and the data rearrangement. An error message including the reason why the arrangement could not be made is created (step ER1).

  Next, the main controller C2 executes detection processing for a management terminal device that manages the storage device (step ER2). Subsequently, the main controller C2 transmits the error message created in step ER1 to the detected terminal device (step ER3).

  Thereafter, the main controller C2 excludes extents that could not be rearranged from extent optimization targets (step ER4). By executing step ER4, the error processing ends.

  In step ER3, the transmission destination is a management terminal device, but the transmission destination is not limited to this. In the present embodiment, the management server of the system may be the transmission destination. Further, in this embodiment, when an error message is acquired, the administrator of the storage apparatus can determine whether it is within an allowable range by referring to the load information of extents that could not be relocated. If the administrator determines that the value is within the allowable range, the storage device can continue the processing as it is.

  In addition, since the extent that could not be rearranged is excluded from the optimization target in step ER4, the optimization process is executed again unless it is newly added to the optimization target manually. No error is output. Therefore, if the administrator determines that relocation is necessary, the administrator needs to change the policy so that relocation is possible, or take measures to reduce the load by adding HDDs. is there.

  Further, the program in the embodiment of the present invention may be a program that causes a computer to execute steps F1 to F9 shown in FIG. 10 and steps F10 to F24 shown in FIG. By installing and executing this program on a computer, the disk array control unit DE-n and the disk array control method in the present embodiment can be realized. Further, the program may further be capable of executing steps ER1 to ER4 shown in FIG.

  When the above program is executed by a computer, the CPU (Central Processing Unit) of the computer functions as the host controller C1, the main controller C2, and the disk controller C3 as described above (see FIG. 4). Further, the memory provided in the computer functions as the cache memory C4 and the memory C5.

  As can be seen from the above description, according to the present embodiment, it is possible to distribute the load on each area of the HDD while suppressing the deterioration of the fault tolerance of the data and the pressure on the processing of the business to be prioritized. As a result, the performance of the storage apparatus is improved. This is because, as described above, in the present embodiment, the transfer destination array and further the transfer destination extent are specified using the policy correspondence table.

  That is, based on the policy correspondence table, the logical volume in which the data to be migrated is stored at the time of data relocation is compared with the logical volume of the migration destination. Then, it is determined whether or not the fault tolerance and performance of the migration destination area required for the logical volume storing the migration source data are satisfied after the migration. For this reason, it is possible to set the relocation destination of the data to be optimized more finely than in the case where the data is optimized based only on the load or the use frequency. As a result, the above-described effect can be obtained.

  In the present embodiment, since a policy is set for each logical volume, it is possible to perform appropriate data optimization according to the importance for each business. This is because, in general, data used in each business is often divided in units of logical volumes according to the business content or its importance.

(Modification)
Here, in the processing shown in FIG. 10 or FIG. 11, a modified example of processing in the storage apparatus when the processing ends without transferring (rearranging) data will be described using FIG. 13 and FIG. 14. To do. FIG. 13 is a diagram for explaining a modification of the embodiment of the present invention.

  As shown in FIG. 13, in this modification, the array includes a hot spare disk HS-1 that is used when a failure occurs in the HDD 100 constituting the array. The hot spare disk is a spare disk, and functions as a substitute for the failed HDD when a failure occurs in any of the HDDs constituting the array having the RAID configuration. By using the hot spare disk, it is possible to prevent the performance degradation of the RAID configuration and to ensure the redundancy.

  Usually, the hot spare disk is held in a state where it is not used at all for data writing or the like in the storage apparatus when no failure has occurred in the RAID configuration. In other words, no I / O load is generated in the storage area of the hot spare disk.

  On the other hand, in this modification, the storage area of the hot spare disk is used as a relocation destination when data relocation cannot be executed. In other words, the extent set in the storage area of the hot spare disk is included in the extents in the array that can be selected as the relocation destination, and the transfer destination candidate extent is selected from these extents. For this reason, it is possible to reliably prevent a delay in processing due to a high load even in an operation where data relocation is not permitted due to low priority.

  Further, in this modification, a hot spare disk that is not often used can be used normally, so that the resources of the entire storage apparatus can be effectively used. However, as shown in FIG. 13, since the extent data composed of a plurality of extent chunks is stored in one HDD constituting the hot spare disk, the storage area spanning the plurality of HDDs is one. Collected in HDD. Therefore, in this case, the load cannot be distributed.

  However, since the data is rearranged in an area where there is no load, the performance in the area grouped in the hot spare disk is improved as compared with the area where the processing delay originally occurred due to high load.

  Note that the relocation of data to the storage area of the hot spare disk is preferably not temporary but temporary. This is because all data is stored in the same HDD, so that data redundancy is completely eliminated and fault tolerance is reduced.

  For this reason, when it is highly necessary to provide redundancy to prevent loss of data, it is preferable to avoid relocation of data to the storage area of the hot spare disk as much as possible. Therefore, the relocation of data to the storage area of the hot spare disk cannot be relocated to the storage area of another array, or the load on the transfer source extent is too high and exceeds the allowable value range. Good to be executed.

  In addition, as an index of an allowable value range, for example, a response time in an extent with a high load can be cited. In addition, when relocation of data to the storage area of the hot spare disk is executed, the administrator needs to consider modification of the design of the logical volume or addition of HDDs constituting the array.

  Here, the processing in the storage apparatus when the processing is completed after the data is rearranged in the storage area of the hot spare disk will be described with reference to FIG. FIG. 14 is a flowchart showing error processing executed by the disk array control unit constituting the storage apparatus in the modification of the embodiment of the present invention.

  As shown in FIG. 14, since the data is rearranged in the storage area of the hot spare disk, first, in the disk array control unit DE-n, the main controller C2 shows an error message indicating that the data has been rearranged in the HS. Is created (step ER11).

  Next, the main controller C2 executes a process of detecting a management terminal device that manages the storage device (step ER12). Subsequently, the main controller C2 transmits the error message created in step ER11 to the detected terminal device (step ER13). Steps ER12 and ER13 are the same steps as steps ER2 and ER3 shown in FIG. 12, respectively. Execution of step ER13 ends the error processing.

  Even when data is rearranged in the storage area of the hot spare disk by executing steps ER11 to ER13, an error message can be sent to the terminal device for the administrator in the same way as a normal error, and the status is notified to the administrator. I can tell you. As a result, the administrator can be warned to relocate the data rearranged in the storage area of the hot spare disk to the storage area of another array.

  In the present embodiment including the above-described modification, it is determined whether the extent (or extent chunk) load is high based on the I / O existence rate disclosed in Patent Document 1. However, the present invention is not limited to this. In the present invention, as a criterion for determining whether the load is high, for example, a simple I / O occurrence rate in an extent or HDD, specifically, an access frequency or the like can be used. Further, the policy correspondence table used in the present invention can be applied even when the optimization of data in units of logical volumes, which is often performed conventionally, is executed.

  The present invention is useful for a storage apparatus that optimizes the data arrangement of a disk array constituting a RAID. According to the present invention, it is possible to optimize the data arrangement in a state where the data policy is satisfied, thereby improving the fault tolerance of the data and suppressing the deterioration of the performance.

1 Storage system 10 Host device 20 Network 100 HDD
A-1, A-2, ... A-i Disk array (array)
S-1, S-2, ... Sn Storage device DE-1, DE-2, ... DE-n Disk array controller C1 Host controller C2 Main controller C3 Disk controller C4 Cache memory C5 Memory D1 Input Module D2 Autonomous optimization control engine D3 I / O information collection module D4 Chunk correspondence table D5 Extent correspondence table D6 Policy correspondence table D7 RAID correspondence table D8 Extent relocation execution module H-1 Database server H-2 File server H-3 File Server HS-1 Hot spare disk L-1, L-2, L-3, L-4, L-5, L-6, L-7 Disk array (array)

Claims (32)

A storage device in which multiple logical volumes that can be recognized from the outside are constructed,
Multiple disk arrays,
A disk array that physically divides a storage area of each of the plurality of disk arrays into a plurality of extents, assigns one or more extents to any of the plurality of logical volumes, and manages the storage areas A control unit,
When the extent of any extent does not satisfy the set condition, the disk array control unit specifies the logical volume to which the transfer source extent is assigned, using the extent as the transfer source extent,
Then, the policy table in which a policy for each of the plurality of logical volumes is registered is compared with the specified logical volume to specify the policy,
Based on the identified policy, a disk array that can be selected as a transfer destination of data stored in the transfer source extent is detected,
The policy for each of the plurality of logical volumes exists in the same disk array as the configuration of the disk array capable of constructing the logical volume, the priority of processing in the data stored in the logical volume, and the logical volume. Including identifiers of other logical volumes that are not allowed,
A storage device. Each of the plurality of disk arrays has a RAID configuration;
The policy includes a RAID configuration capable of constructing the logical volume as a configuration of the disk array capable of constructing the logical volume,
The disk array control unit can further select each of the plurality of disk arrays using a RAID correspondence table in which the type of RAID configuration of the disk array and the performance of the disk array are registered. The storage apparatus according to claim 1, wherein the storage apparatus detects a disk array. The policy further includes, for each of the plurality of logical volumes, an index indicating the performance of a disk array capable of constructing the logical volume,
The disk array control unit further includes:
From the extents in the selectable disk array, select an extent satisfying a load setting condition as a transfer destination candidate extent,
Using the policy table, specify the policy of the logical volume to which the transfer destination candidate extent is allocated,
The data of the transfer source extent can be transferred to the transfer destination candidate extent by comparing the policy of the logical volume to which the transfer destination candidate extent is assigned with the policy of the logical volume to which the transfer source extent is assigned. The storage apparatus according to claim 1, wherein the storage apparatus determines whether or not. The policy further includes, for each of the plurality of logical volumes, an allowable load that the disk array in which the logical volume is constructed can allow the other logical volumes;
When the disk array control unit further determines that transfer to the transfer destination candidate extent is possible,
Identify the logical volume policy assigned to other extents in the disk array containing the transfer destination candidate extents,
The policy of the logical volume to which the other extent is allocated is compared with the load of the transfer source extent, and it is determined again whether the data of the transfer source extent can be transferred to the transfer destination candidate extent. The storage device according to claim 3.   The disk array control unit notifies the outside that the data cannot be transferred when it is determined that the data of the transfer source extent cannot be transferred to the transfer destination candidate extent. The storage apparatus according to 3 or 4. All or a part of the plurality of disk arrays includes a spare disk,
The selectable disk including an extent set in the storage area of the spare disk when the disk array control unit detects a disk array having the spare disk as the selectable disk array. The storage apparatus according to any one of claims 3 to 5, wherein the transfer destination candidate extent is selected from extents in the array. The disk array control unit selects an extent set in the storage area of the spare disk as the transfer destination candidate extent, and sets the data of the transfer source extent as an extent set in the storage area of the spare disk. When it is determined that transfer is possible, and then the data of the transfer source extent is transferred to an extent set in the storage area of the spare disk,
The storage device according to claim 6, which notifies the outside that the spare disk is being used. When the storage device is connected to another storage device having a plurality of disk arrays via a network,
The disk array control unit detects the selectable disk array from the extent set in the disk array included in the storage device and the extent set in the disk array included in the other storage device; The storage device according to any one of claims 1 to 7. A host device and a storage device in which a plurality of logical volumes recognizable from the host device are constructed,
The storage device physically divides a plurality of disk arrays and a storage area of each of the plurality of disk arrays into a plurality of extents, and allocates one or more extents to any one of the plurality of logical volumes. And a disk array control unit for managing the storage area,
When the extent of any extent does not satisfy the set condition, the disk array control unit specifies the logical volume to which the transfer source extent is assigned, using the extent as the transfer source extent,
Then, the policy table in which a policy for each of the plurality of logical volumes is registered is compared with the specified logical volume to specify the policy,
Based on the identified policy, a disk array that can be selected as a transfer destination of data stored in the transfer source extent is detected,
The policy for each of the plurality of logical volumes includes at least the configuration of a disk array that can construct the logical volume, the priority of processing in the data stored in the logical volume, and the same disk array as the logical volume. Including identifiers of other logical volumes that are not allowed to exist,
A storage system characterized by that. Each of the plurality of disk arrays has a RAID configuration;
The policy includes a RAID configuration capable of constructing the logical volume as a configuration of the disk array capable of constructing the logical volume,
The disk array control unit can further select each of the plurality of disk arrays using a RAID correspondence table in which the type of RAID configuration of the disk array and the performance of the disk array are registered. The storage system according to claim 9, wherein a disk array is detected. The policy further includes, for each of the plurality of logical volumes, an index indicating the performance of a disk array capable of constructing the logical volume,
The disk array control unit further includes:
From the extents in the selectable disk array, select an extent satisfying a load setting condition as a transfer destination candidate extent,
Using the policy table, specify the policy of the logical volume to which the transfer destination candidate extent is allocated,
The data of the transfer source extent can be transferred to the transfer destination candidate extent by comparing the policy of the logical volume to which the transfer destination candidate extent is assigned with the policy of the logical volume to which the transfer source extent is assigned. The storage system according to claim 9 or 10, wherein it is determined whether or not. The policy further includes, for each of the plurality of logical volumes, an allowable load that the disk array in which the logical volume is constructed can allow the other logical volumes;
When the disk array control unit further determines that transfer to the transfer destination candidate extent is possible,
Identify the logical volume policy assigned to other extents in the disk array containing the transfer destination candidate extents,
The policy of the logical volume to which the other extent is allocated is compared with the load of the transfer source extent, and it is determined again whether the data of the transfer source extent can be transferred to the transfer destination candidate extent. The storage system according to claim 11.   The disk array control unit notifies the outside that the data cannot be transferred when it is determined that the data of the transfer source extent cannot be transferred to the transfer destination candidate extent. The storage system according to 11 or 12. All or a part of the plurality of disk arrays includes a spare disk,
The selectable disk including an extent set in the storage area of the spare disk when the disk array control unit detects a disk array having the spare disk as the selectable disk array. The storage system according to any one of claims 11 to 13, wherein the transfer destination candidate extent is selected from the extents in the array. The disk array control unit selects an extent set in the storage area of the spare disk as the transfer destination candidate extent, and sets the data of the transfer source extent as an extent set in the storage area of the spare disk. When it is determined that transfer is possible, and then the data of the transfer source extent is transferred to an extent set in the storage area of the spare disk,
The storage system according to claim 14, wherein the storage system notifies the host device that the spare disk is being used. A plurality of the storage devices are provided, and each of the storage devices is connected to each other via a network,
In each of the storage devices, the disk array control unit includes the extent set in the disk array included in the storage device provided with the storage device and the extent set in the disk array included in the other storage device. The storage system according to any one of claims 9 to 15, wherein a selectable disk array is detected. In a storage apparatus in which a plurality of logical volumes recognizable from the outside are constructed, a method for controlling a plurality of disk arrays provided therein,
(A) physically dividing a storage area of each of the plurality of disk arrays into a plurality of extents, and allocating one or two or more extents to any of the plurality of logical volumes;
(B) when the load of any extent does not satisfy the set condition, the logical volume to which the transfer source extent is assigned is determined using the extent as the transfer source extent; and
(C) For each of the plurality of logical volumes, the configuration of the disk array capable of constructing the logical volume, the priority of processing in the data stored in the logical volume, and the disk array that is the same as the logical volume Collating the logical volume identified in step (b) with a policy table in which a policy is registered, including identifiers of other logical volumes that are not allowed to be identified, and identifying the policy; and
And (d) detecting a disk array that can be selected as a transfer destination of data stored in the transfer source extent based on the policy specified in the step (c). Disk array control method. Each of the plurality of disk arrays has a RAID configuration;
In the step (c), the policy includes a RAID configuration capable of constructing the logical volume as a configuration of a disk array capable of constructing the logical volume, and
In the step (d), for each of the plurality of disk arrays, the selectable disk can be selected using a RAID correspondence table in which the type of RAID configuration of the disk array and the performance of the disk array are registered. The disk array control method according to claim 17, wherein the array is detected. In the step (c), the policy further includes, for each of the plurality of logical volumes, an index indicating the performance of the disk array capable of constructing the logical volume, and (e) in the selectable disk array Selecting an extent satisfying a set load condition as a transfer destination candidate extent from among the extents of the file, and using the policy table, identifying a policy of a logical volume to which the transfer destination candidate extent is allocated, When,
(F) By comparing the policy of the logical volume to which the transfer destination candidate extent is allocated with the policy of the logical volume to which the transfer source extent is allocated, the data of the transfer source extent is converted to the transfer destination candidate extent. 19. The disk array control method according to claim 17 or 18, further comprising the step of determining whether or not transfer is possible. In the step (c), the policy further includes, for each of the plurality of logical volumes, an allowable load that the disk array in which the logical volume is constructed can permit the other logical volumes. ,
(G) When it is determined in the step (f) that transfer to the transfer destination candidate extent is possible, the policy of the logical volume allocated to other extents in the disk array including the transfer destination candidate extent is changed. Identify, step,
(H) After the execution of the step (g), the policy of the logical volume to which the other extent is allocated is compared with the load of the transfer source extent, and the data of the transfer source extent is stored again. The disk array control method according to claim 19, further comprising the step of determining whether or not transfer to a transfer destination candidate extent is possible. (I) When it is determined that the data of the transfer source extent cannot be transferred to the transfer destination candidate extent, the step of notifying the outside that the data cannot be transferred is further included. The disk array control method according to claim 19 or 20. All or a part of the plurality of disk arrays includes a spare disk,
When a disk array having the spare disk is detected as the selectable disk array in the step (d), the extent set in the storage area of the spare disk in the step (e) The disk array control method according to any one of claims 19 to 21, wherein the transfer destination candidate extent is selected from among the extents in the selectable disk array including. (J) In the step (e), an extent set in the storage area of the spare disk is selected as the transfer destination candidate extent, and in the step (f), the data of the transfer source extent is selected. When it is determined that the data can be transferred to the extent set in the storage area of the spare disk, and then the data of the transfer source extent is transferred to the extent set in the storage area of the spare disk,
23. The disk array control method according to claim 22, further comprising a step of notifying the outside that the spare disk is being used. When there are a plurality of the storage devices and each of the storage devices is connected to each other via a network,
The disk array control according to any one of claims 17 to 23, wherein in the step (d), the selectable disk array is detected from extents set in the disk arrays included in the plurality of storage devices. Method. A program for controlling, by a computer, a plurality of disk arrays in which a plurality of logical volumes recognizable from the outside are constructed,
In the computer,
(A) physically dividing a storage area of each of the plurality of disk arrays into a plurality of extents, and allocating one or two or more extents to any of the plurality of logical volumes;
(B) when the load of any extent does not satisfy the set condition, the logical volume to which the transfer source extent is allocated is specified using the extent as the transfer source extent; and
(C) For each of the plurality of logical volumes, the configuration of the disk array capable of constructing the logical volume, the priority of processing in the data stored in the logical volume, and the disk array that is the same as the logical volume Collating the logical volume identified in step (b) with a policy table in which a policy is registered, including identifiers of other logical volumes that are not allowed to be identified, and identifying the policy; and
And (d) detecting a disk array that can be selected as a transfer destination of data stored in the transfer source extent based on the policy specified in the step (c). Program to do. Each of the plurality of disk arrays has a RAID configuration;
In the step (c), the policy includes a RAID configuration capable of constructing the logical volume as a configuration of a disk array capable of constructing the logical volume, and
In the step (d), for each of the plurality of disk arrays, the selectable disk can be selected using a RAID correspondence table in which the type of RAID configuration of the disk array and the performance of the disk array are registered. The program according to claim 25, wherein the program detects an array. In the step (c), the policy further includes, for each of the plurality of logical volumes, an index indicating the performance of the disk array capable of constructing the logical volume, and (e) in the selectable disk array Selecting an extent satisfying a set load condition as a transfer destination candidate extent from among the extents of the file, and using the policy table, identifying a policy of a logical volume to which the transfer destination candidate extent is allocated, When,
(F) By comparing the policy of the logical volume to which the transfer destination candidate extent is allocated with the policy of the logical volume to which the transfer source extent is allocated, the data of the transfer source extent is converted to the transfer destination candidate extent. 27. The program according to claim 25 or 26, further comprising: causing the computer to execute a step of determining whether or not transfer is possible. In the step (c), the policy further includes, for each of the plurality of logical volumes, an allowable load that the disk array in which the logical volume is constructed can permit the other logical volumes. ,
(G) When it is determined in the step (f) that transfer to the transfer destination candidate extent is possible, the policy of the logical volume allocated to other extents in the disk array including the transfer destination candidate extent is changed. Identify, step,
(H) After the execution of the step (g), the policy of the logical volume to which the other extent is allocated is compared with the load of the transfer source extent, and the data of the transfer source extent is stored again. 28. The program according to claim 27, further causing the computer to execute a step of determining whether or not transfer to a transfer destination candidate extent is possible. (I) when it is determined that the data of the transfer source extent cannot be transferred to the transfer destination candidate extent, the step of notifying the outside that the data cannot be transferred is further included in the computer The program according to claim 27 or 28, wherein the program is executed. All or a part of the plurality of disk arrays includes a spare disk,
When a disk array having the spare disk is detected as the selectable disk array in the step (d), the extent set in the storage area of the spare disk in the step (e) 30. The program according to any one of claims 27 to 29, wherein the transfer destination candidate extent is selected from the extents in the selectable disk array. (J) In the step (e), an extent set in the storage area of the spare disk is selected as the transfer destination candidate extent, and in the step (f), the data of the transfer source extent is selected. When it is determined that the data can be transferred to the extent set in the storage area of the spare disk, and then the data of the transfer source extent is transferred to the extent set in the storage area of the spare disk,
The program according to claim 30, further causing the computer to further execute a step of notifying that the spare disk is being used. When there are a plurality of the storage devices and each of the storage devices is connected to each other via a network,
The program according to any one of claims 25 to 31, wherein in the step (d), the selectable disk array is detected from extents set in the disk arrays included in the plurality of storage devices.

Images